Finding Almost Squares II

Tsz Ho Chan

November 27, 2006

Abstract
In this article, we study short intervals that contain another type of “almost square”, an integer n   which can be factored in two different ways n = a 1 b 1 = a 2 b 2   with a 1 , a 2 , b 1 , b 2   close to n   .

1 Introduction

In [ 1 ], the author studied the problem of finding “almost squares” in short intervals, namely:
Question 1. For 0 ≤ θ < 1 / 2   , what is the least f ( θ )   such that, for some c 1 , c 2 > 0   , any interval [ x − c 1 x f ( θ ) , x + c 1 x f ( θ ) ]   contains an integer n   with n = a b   , where a   , b   are integers in the interval [ x 1 / 2 − c 2 x θ , x 1 / 2 + c 2 x θ ]   ? Note: c 1   and c 2   may depend on θ   .
A similar question is the following:
Question 2. For 0 ≤ θ < 1 / 2   , what is the least g ( θ )   such that, for some c 1 , c 2 > 0   , any interval [ x − c 1 x g ( θ ) , x + c 1 x g ( θ ) ]   contains an integer n   with n = a 1 b 1 = a 2 b 2   , where a 1 < a 2 ≤ b 2 < b 1   are integers in the interval [ x 1 / 2 − c 2 x θ , x 1 / 2 + c 2 x θ ]   ? Note: c 1   and c 2   may depend on θ   .
Note: Actually, the author first considered Question  2 and then turned to Question  1 which has connection to the problem on the distribution of n 2 α ( m o d 1 )   and the problem on gaps between sums of two squares.
In [ 1 ], we showed that f ( θ ) = 1 / 2   when 0 ≤ θ < 1 / 4   , f ( 1 / 4 ) = 1 / 4   and f ( θ ) ≥ 1 / 2 − θ   . We conjectured that f ( θ ) = 1 / 2 − θ   for 1 / 4 < θ < 1 / 2   and gave conditional result when 1 / 4 < θ < 3 / 10   . To Question  2 , we have the following
Theorem 1.1. For 0 < θ < 1 / 4   , g ( θ )   does not exist (i.e. all possible products of pairs of integers in [ x 1 / 2 − c 2 x θ , x 1 / 2 + c 2 x θ ]   are necessarily distinct for large x   ).
Theorem 1.2. For 1 / 4 ≤ θ < 1 / 2   , g ( θ ) ≥ 1 − 2 θ   .
Theorem 1.3. For 1 / 4 ≤ θ ≤ 1 / 3   , g ( θ ) ≤ 1 − θ   .
We believe that the lower bound is closer to the truth and conjecture
Conjecture 1.1. For 1 / 4 ≤ θ < 1 / 2   , g ( θ ) = 1 − 2 θ   .

2 Preliminaries and 0 ≤ θ < 1 / 4  

Suppose n = a 1 b 1 = a 2 b 2   with x 1 / 2 − c 2 x θ ≤ a 1 < a 2 ≤ b 2 < b 1 ≤ x 1 / 2 + c 2 x θ   .
Let d 1 = ( a 1 , a 2 )   and d 2 = ( b 1 , b 2 )   be the greatest common divisors. Then we must have d 1 , d 2 > 1   . For otherwise, say d 1 = 1   , then a 2   divides b 1   which implies x 1 / 2 + c 2 x θ ≥ b 1 ≥ 2 a 2 ≥ 2 x 1 / 2 − 2 c 2 x θ   . This is impossible for large x   as θ < 1 / 2   . Now, let a 1 = d 1 e 1   , a 2 = d 1 e 2   , b 1 = d 2 f 1   and b 2 = d 2 f 2   . Here ( e 1 , e 2 ) = 1 = ( f 1 , f 2 )   . Then n = d 1 e 1 d 2 f 1 = d 1 e 2 d 2 f 2 gives e 1 f 1 = e 2 f 2 .   Due to co-primality, e 2 = f 1   and e 1 = f 2   . Therefore,
n = ( d 1 e 1 ) ( d 2 e 2 ) = ( d 1 e 2 ) ( d 2 e 1 ) (1)
with 1 < d 1 < d 2   , e 1 < e 2   and ( e 1 , e 2 ) = 1   .
Now, from a 2 − a 1 ≤ 2 c 2 x θ   , d 1 ≤ d 1 e 2 − d 1 e 1 ≤ 2 c 2 x θ   . Similarly, one can deduce that d 2 , e 1 , e 2 ≤ 2 c 2 x θ   . Moreover, as d 1 e 1 = a 1 ≥ x 1 / 2 − c 2 x θ   , we have d 1 , e 1 ≥ 1 2 c 2 x 1 / 2 − θ − 1 2   . Similarly, d 2 , e 2 ≥ 1 2 c 2 x 1 / 2 − θ − 1 2   . Summing up, we have
1 2 c 2 x 1 / 2 − θ − 1 2 ≤ d 1 , d 2 , e 1 , e 2 ≤ 2 c 2 x θ . (2)
From ( 2 ), we see that no such n   exists for 0 ≤ θ < 1 / 4   and hence Theorem  1.1 .

3 Lower bound for g ( θ )  

From ( 1 ) and ( 2 ), we see that an integer n = a 1 b 1 = a 2 b 2   , satisfying the conditions for a 1 , a 2 , b 1 , b 2   in Question  2 , must be of the form:
n = ( d 1 e 1 ) ( d 2 e 2 ) with 1 2 c 2 x 1 / 2 − θ − 1 2 ≤ d 1 , d 2 , e 1 , e 2 ≤ 2 c 2 x θ   and x 1 / 2 − c 2 x θ ≤ d 1 e 1 < d 1 e 2 , d 2 e 1 < d 2 e 2 ≤ x 1 / 2 + c 2 x θ   . In particular, e 2 d 2 − e 2 d 1 ≤ 2 c 2 x θ   which implies e 2 − e 1 ≤ 2 c 2 x θ / d 2   . Similarly, d 2 − d 1 ≤ 2 c 2 x θ / e 2   . Thus, the number of such quartuple ( d 1 , d 2 , e 1 , e 2 )   is bounded by ≪ ∑ x 1 / 2 − θ ≪ d 2 , e 2 ≪ x θ x 1 / 2 − c 2 x θ ≤ d 2 e 2 ≤ x 1 / 2 + c 2 x θ x θ e 2 x θ d 2 ≪ x 2 θ x 1 / 2 x θ x ε = x 3 θ − 1 / 2 + ε   for any ε > 0   as the number of divisor function d ( n ) ≪ n ε   . It follows that there are at most O ( x 3 θ − 1 / 2 + ε )   such integers n   in the interval [ x − c 2 x 1 / 2 + θ / 3 , x + c 2 x 1 / 2 + θ / 3 ]   . Therefore, some two consecutive such n   's have gap ≫ x 1 / 2 + θ x 3 θ − 1 / 2 + ε = x 1 − 2 θ − ε .   Pick y   to be the midpoint between these two integers. Then, for some constant c > 0   , the interval [ y − c y 1 − 2 θ − ε , y + c y 1 − 2 θ − ε ]   does not contain any integer n = a 1 b 1 = a 2 b 2   with y 1 / 2 − c 2 y θ / 2 ≤ a 1 < a 2 ≤ b 2 < b 1 ≤ y 1 / 2 + c 2 y θ / 2   as x − c 2 x 1 / 2 + θ / 3 ≤ y ≤ x + c 2 x 1 / 2 + θ / 3   . Consequently, for any constant c , c ′ > 0   , there is arbitrarily large y   such that the interval [ y − c y 1 − 2 θ − 2 ε , y + c y 1 − 2 θ − 2 ε ]   does not contain any integer n = a 1 b 1 = a 2 b 2   with y 1 / 2 − c ′ y θ ≤ a 1 < a 2 ≤ b 2 < b 1 ≤ y 1 / 2 + c ′ y θ   . Therefore, g ( θ ) ≥ 1 − 2 θ − 2 ε   which gives Theorem  1.2 by letting ε → 0   .

4 Upper bound for g ( θ )  

Proof of Theorem  1.3 : For any large x   , set N = [ x 1 / 4 ]   and ξ = { x 1 / 4 }   , the integer part and fractional part of x 1 / 4   respectively. Based on ( 1 ), we are going to pick, for 0 ≤ ε ≤ 1 / 2   ,
d 1 = q N + r 1 , d 2 = q N + r 2 , e 1 = N + s 1 q , e 2 = N + s 2 q (3)
for some 1 ≤ q ≤ N ε   , 0 ≤ r 1 , r 2 < N   and s 1 , s 2 ≪ q   with N ≡ − s 1 ≡ − s 2 ( m o d q )   . Our goal is to make x = ( N + ξ ) 4 = N 4 + 4 N 3 ξ + O ( N 2 ) ≈ ( q N + r 1 ) N + s 1 q ( q N + r 2 ) N + s 2 q .   The right hand side above is
= [ N 2 + ( r 1 q + s 1 ) N + r 1 s 1 q ] [ N 2 + ( r 2 q + s 2 ) N + r 2 s 2 q ] = N 4 + ( r 1 + r 2 q + s 1 + s 2 ) N 3 + [ r 1 s 1 q + r 2 s 2 q + ( r 1 q + s 1 ) ( r 2 q + s 2 ) ] N 2 + [ r 1 s 1 q ( r 2 q + s 2 ) + r 2 s 2 q ( r 1 q + s 1 ) ] N + r 1 s 1 r 2 s 2 q 2 (4)
By Dirichlet's Theorem on diophantine approximation, we can find integer 1 ≤ q ≤ N ε   such that | 4 ξ − p q | ≤ 1 q N ε   for some integer p   . Fix such a q   . Then, pick s 1 < s 2 < 0   to be the largest two integers such that N ≡ − s 1 ≡ − s 2 ( m o d q )   . Clearly, s 1 , s 2 ≪ q   . Then, one simply picks some 0 < r 1 < r 2 ≪ q 2   such that r 1 + r 2 q + s 1 + s 2 = p q   . With these values for q , r 1 , r 2 , s 1 , s 2   , ( 4 ) is = N 4 + 4 N 3 ξ + O ( N 3 − ε ) + O ( q 2 N 2 ) + O ( q 3 N ) + O ( q 4 ) .   Hence, we have just constructed an integer n = d 1 e 1 d 2 e 2   which is within O ( N 3 − ε ) + O ( N 2 + 2 ε ) = O ( x 3 / 4 − ε / 4 ) + O ( x 1 / 2 + ε / 2 ) = O ( x 3 / 4 − ε / 4 )   from x   if ε ≤ 1 / 3   . One can easily check that a 1 = d 1 e 1   , b 1 = d 2 e 2   , a 2 = d 1 e 2   and b 2 = d 2 e 1   are in the interval [ x 1 / 2 − C x 1 / 4 + ε / 4 , x 1 / 2 + C x 1 / 4 + ε / 4 ]   for some constant C > 0   . Set θ = 1 / 4 + ε / 4   . We have, for some C ′ > 0   , n = a 1 b 1 = a 2 b 2   in the interval [ x − C ′ x 1 − θ , x + C ′ x 1 − θ ]   such that a 1 < a 2 , b 2 < b 1   are integers in [ x 1 / 2 − C x θ , x 1 / 2 + C x θ ]   provided 1 / 4 ≤ θ ≤ 1 / 4 + 1 / 12 = 1 / 3   . This proves Theorem  1.3 .

5 Open questions

Conjecture  1.1 may be too hard to prove in the moment. As a starting point, can one show that g ( 1 / 4 ) = 1 / 2   ? Or even g ( 1 / 4 ) < 3 / 4   ? Another possibility would be trying to get some results conditionally like [ 1 ]. Also, one may consider g ( θ )   when θ   is near to 1 / 2   . This leads to the problem about gaps between integers that have more than one representation as a sum of two squares.
Acknowledgement The author would like to thank the American Institute of Mathematics for providing a stimulating environment to work at.
References

  1. Tsz Ho Chan, Finding Almost Squares, preprint, 2005, arXiv:math.NT/0502199.

Tsz Ho Chan American Institute of Mathematics 360 Portage Avenue Palo Alto, CA 94306 U.S.A. thchan@aimath.org